Queensland Health Payroll

What Happened

Queensland Health is one of the largest employers in Australia, responsible at the time for paying roughly 78,000 staff — nurses, doctors, allied health workers and administrative employees — spread across hundreds of hospitals and facilities in the most geographically dispersed state in the country. Its payroll was not a simple salary run. Pay was governed by 13 industrial awards and multiple agreements producing, by evidence given to the Commission of Inquiry, more than 24,000 different pay combinations. By the mid-2000s the payroll ran on a system called LATTICE that was more than a decade old, and the vendor had advised that support would end in July 2008. The origins of the disaster lie one level up, in a whole-of-government scheme. In the mid-2000s Queensland Treasury pursued a Shared Services Initiative intended to standardise HR and finance systems across all departments on SAP software. Progress was slow and expensive; a former Accenture managing director later told the Commission that between 2005 and 2007 the government was, in his words, burning through its budget without commensurate outcomes. Reviews, including one known as the Kelliher Report, found the programme significantly behind schedule. The government therefore decided to hand systems integration to a single prime contractor. In December 2007, after a procurement process the Commission would later find was not administered properly, IBM Australia was appointed prime contractor. The contract to design and implement the Queensland Health payroll system was fixed at A$6.19 million, with delivery promised by July 2008 — the same month LATTICE support expired. As the Commission's report records, IBM had agreed to deliver, at a fixed price, a project for which no settled statement of work existed and against requirements the departments themselves were still debating. The state's own legal advisers had rated IBM's offer as raising more material issues than competing bids. The project unravelled almost immediately. Scope churned, costs escalated through contract variations, and by October 2008 the whole-of-government rollout was abandoned and IBM's contract was rescoped down to Queensland Health alone. By the time the system went live, the amount paid to IBM had exceeded A$37 million, and the government's own implementation costs added a further A$64 million. Delivery slipped by roughly twenty months past the original July 2008 date. Go-live itself became a rolling crisis. There were, by the account of academic analyses of the inquiry evidence, ten failed attempts before the system was finally switched on. User acceptance testing had identified serious defects; the decision to proceed on 14 March 2010 was taken under intense time pressure, with a rescue option — reverting to the old arrangement — no longer realistically available because LATTICE support had ended. The Commission examined that go-live decision in detail and was critical of how it was made. The consequences were immediate and human. The system produced roughly 35,000 payroll anomalies. Thousands of staff were underpaid or received nothing at all; some payslips read zero. Others were overpaid, in some cases substantially, and some of those overpaid were initially accused of fraud. A system intended to be efficient, economical and largely automated instead required more than 1,000 payroll staff performing about 200,000 manual operations and processing 92,000 forms every fortnight, according to the KPMG figures cited in the Commission's report. The financial reckoning came in stages. The Auditor-General of Queensland reported to Parliament in 2010 (Report No. 7). KPMG's May 2012 review recorded that operating costs to that date had exceeded A$400 million, with an estimated A$385 million more required over the following five years. Queensland Government estimates of the total cost of ownership — development, remediation and years of elevated operating cost — were subsequently put at around A$1.2 billion, a figure widely reported as approximately A$1.25 billion; this is an estimate, attributed to those reviews, not a single audited number. Accountability followed a tortuous path. The Bligh government reached a settlement with IBM in 2010 that released the company from further claims, fearing — as later reporting described — a counterclaim and the loss of IBM's support for a barely functioning system. On 13 December 2012 the Newman government ordered a Commission of Inquiry under the Honourable Richard Chesterman QC. His report, delivered in July 2013, described the episode as one that must take its place "in the front rank of failures in public administration in this country". The Commission found the procurement process flawed and made adverse findings about the conduct of particular individuals in the evaluation process — while being precise, as this article must be, that it examined whether laws had been broken and did not find that the failure was the product of criminal corruption across the board. The government banned IBM from new state contracts. The payroll system, stabilised only with hundreds of additional staff, continued to pay Queensland Health's workforce at an extraordinary running cost.

The Promise

The promise was seductively modest, and that was part of the danger. Queensland Health would replace a payroll system whose vendor support was expiring with a modern, SAP-based solution that would be efficient, economical and largely automated. Payroll for 78,000 staff would flow through standardised processes rather than manual workarounds; award interpretation, allowances, penalties and overtime would be calculated by software rather than by armies of payroll clerks. The fixed price — A$6.19 million — implied that the problem was well understood and bounded: an off-the-shelf product, configured by an experienced prime contractor, delivered in about seven months.

At the whole-of-government level, the promise was grander: a shared services model in which every Queensland department would eventually run common HR and finance systems, harvesting economies of scale and ending the duplication of dozens of agency-specific platforms. Ministers were told to expect substantial savings.

Neither promise was interrogated with the scepticism a fixed price of that size demanded. A payroll covering 13 awards and more than 24,000 pay combinations is not a commodity product; it is one of the most complex rule-sets in Australian public administration. The price signalled not efficiency but misunderstanding — on both sides of the contract — of what was actually being bought.

The Warning Signs

The documented warnings — visible before go-live, not constructed afterwards — were numerous and specific. First, the state's own experience: between 2005 and 2007 the whole-of-government SAP programme had consumed large sums without delivering, and external consultants had warned Treasury in writing and in person that it was not a competent systems implementer. The Kelliher review found the programme significantly behind schedule before IBM was ever engaged.

Second, the procurement itself. The Commission found the evaluation process was not administered properly: there was no probity advisor for key stages, no functioning conflicts register, and evaluation criteria shifted during the process. The state's external legal advisers assessed IBM's tender as raising more material issues than rival bids. Departments had not agreed their own requirements — witnesses told the Commission that, on the eve of contract, they were still arguing about what they would accept.

Third, the arithmetic. A seven-month, A$6.19 million fixed price for that rule complexity was, as one analysis put it, a gross underestimation visible to any experienced estimator.

Fourth, the testing evidence. User acceptance testing before March 2010 identified serious defects, and there had already been multiple aborted go-live attempts. These were facts in the hands of decision-makers, not hindsight.

What hindsight adds is mainly confirmation: the scale of the eventual operating cost (A$400 million-plus by 2012) could not have been forecast precisely. But the failure itself was forecast, repeatedly, by the project's own documents.

Governance Analysis

Governance failed at every layer the Project Failure Pyramid describes, and the Commission of Inquiry documented each failure with unusual precision.

At the foundation sat a false premise: that payroll modernisation was a bounded technical purchase rather than a complex business transformation. Because the premise was wrong, everything built on it — the seven-month schedule, the fixed price, the light governance structure — was wrong too. The Decision Quality Model asks whether decision-makers had accurate information, genuine options and the capacity to act; the evidence shows the state's departments had not even agreed their requirements when the contract was signed.

Structure compounded the error. Responsibility was split between Queensland Health (the business owner), CorpTech (the Treasury unit managing the prime contractor) and IBM (the contractor), with no single empowered owner of the outcome. CorpTech was managing a prime contractor for the first time; IBM was serving as a prime contractor on this type of engagement for the first time. Each party could plausibly believe risk sat elsewhere — the classic pre-condition for concealed problems.

Escalation then inverted the normal logic. As costs rose through variations and the schedule collapsed, the governance response was not to stop but to press on, because the old system's support deadline had passed and there was no way back. By the go-live decision, sunk cost and deadline panic had replaced engineering judgement. The Commission was pointedly critical of how that decision was made: known defects were weighed against a deadline, and the deadline won.

Finally, accountability after the fact was weak. The 2010 settlement released IBM before the full cost was known; ministers and senior officials moved on; and it took a change of government and a royal-commission-style inquiry to establish the public record. The Leadership Blind Spot Matrix fits precisely: leaders were insulated from operational reality (payroll clerks drowning in manual work), from contractual reality (what A$6.19 million could actually buy) and from temporal reality (what a seven-month build implied). None of these blind spots required bad faith. They required only a governance structure in which nobody's job depended on finding out the truth early.

The Human Impact

The Commission of Inquiry put the human cost on the record in plain language, and it deserves quoting in substance. When the system went live in March 2010, thousands of Queensland Health staff — the nurses, doctors and support workers running the state's hospitals — were underpaid or received nothing at all. The Commission recorded that a number of employees were made temporarily destitute, unable to afford the basic necessities of life. Some who were overpaid through no fault of their own were falsely accused of fraud, and overpayment recovery efforts pursued staff for years afterwards; as of 2017, reporting indicated tens of thousands of workers still owed millions in overpaid wages the system itself had generated. The Commission described it as, for all affected, a time of great anxiety and hardship. Health Minister Paul Lucas resigned over the affair, and the scandal contributed to the Bligh government's defeat in 2012. No inquiry can measure the quieter damage: staff who left the health system, and patients cared for by people worried about whether they had been paid.

Was It a Failure?

By any honest measure, yes — but the shape of the failure matters. Technically, the system eventually worked: by 2013 the payroll had stabilised, and Queensland Health staff are still paid through descendants of this platform. If the test is "does payroll run?", the project ultimately passed. But that is the wrong test, and setting it would excuse almost anything given enough money and time.

Against every criterion the project set for itself — cost, schedule, automation, efficiency — it failed comprehensively. A A$6.19 million fixed-price contract became more than A$100 million of project cost before go-live, and an estimated A$1.2 billion-plus total cost of ownership (attributed to KPMG and government estimates). A system meant to eliminate manual processing required more than 1,000 additional payroll staff and 200,000 manual operations a fortnight. And the failure was not merely financial: it inflicted documented hardship on the workforce it served.

The Commission's verdict — "the front rank of failures in public administration in this country" — is the considered judgement of the most thorough official examination the project received. It is also worth stating what the failure was not: the Commission did not find it was the product of systemic criminality. It was something more instructive and more common — ordinary institutional dysfunction, compounding.

Why It Matters

The Queensland Health payroll disaster matters beyond Australia because it is the best-documented case in the world of a pattern that recurs wherever governments buy large IT systems: the UK's NHS National Programme for IT, numerous US state ERP failures, and countless quieter write-offs. The Commission of Inquiry created something rare — a complete, sworn, public evidentiary record, from the first treasury business case to the go-live decision. Most failed projects bury their records in confidential settlements; Queensland's are on the public record, which makes this case the canonical teaching text. Three lessons generalise. First, fixed price is not fixed risk: a low fixed price for an undefined scope is simply a dispute deferred, and the customer — not the contractor — held the operational risk all along, because hospitals cannot stop paying nurses. Second, deadlines created by vendor support expiry are governance failures in themselves: Queensland let its negotiating position collapse by running the old system to the end of its life before its replacement existed. Third, the case shows how failure compounds: procurement flaws produced a bad contract, the bad contract produced scope warfare, scope warfare produced a defective system, and the absence of a fallback produced a coerced go-live. Each stage was individually survivable; the chain was not. For Asian governments now digitising health and civil-service payrolls at scale — India, Indonesia, the Philippines, Vietnam — this is the reference case for what not to sign, and for the procurement, probity and testing disciplines that cost little beside A$1.25 billion.